Biggest savings on speech models.
Prices checked
Eleven Multilingual v2 at Vercel AI Gateway leads 8 models at $40 per 1M characters, checked .
Top 8
all speech models- 1
Eleven Multilingual v2
Trains on prompts
Vercel AI Gateway · 50% less than ElevenLabs · 3 providers$40per 1M characters - 2
Eleven Flash v2
Trains on promptsVercel AI Gateway · 50% less than ElevenLabs · 2 providers$20per 1M characters - 3
Eleven Flash v2.5
Trains on promptsVercel AI Gateway · 50% less than ElevenLabs · 2 providers$20per 1M characters - 4
Eleven v3
Trains on promptsVercel AI Gateway · 50% less than ElevenLabs · 2 providers$40per 1M characters - 5
Eleven v3 Conversational
Trains on promptsVercel AI Gateway · 50% less than ElevenLabs · 2 providers$20per 1M characters - 6
Eleven v4
Trains on promptsVercel AI Gateway · 50% less than ElevenLabs · 2 providers$40per 1M characters - 7
Eleven v4 Turbo
Trains on promptsVercel AI Gateway · 50% less than ElevenLabs · 2 providers$20per 1M characters - 8
Scribe v2
Trains on promptsVercel AI Gateway · 50% less than ElevenLabs · 2 providers$0.11per hour of audio
How we count
Every row here is below the lab's own price, with a published reason behind it. A price below the lab's own that nothing accounts for is listed on the model's page and not ranked. 8 models have a reason we can point at, the widest gap being 50%. Reading text aloud is priced per million characters and transcription per hour of audio, and every price below says which.
A price below the lab's own is listed only when the provider says why.
- Trains on prompts
- The provider says it may train on what you send it, which is what pays for the lower price.
Frequently asked
- What is the cheapest of the speech prices below the lab's own right now?
- Scribe v2 at Vercel AI Gateway, $0.11 per 1M characters, read on 31 Oct 2026.
- Which lab shows up most in this ranking?
- ElevenLabs trained 8 of the top 8 models here.
- How is this ranked, and how current is it?
- Every row is the cheapest price for that model that we can hold beside the others: the lab's own, a price below it with a published reason, or a price at or above it. A price that buys a different task, or a resolution the lab's own price does not state, or a unit we will not convert without publishing the assumption, is listed on the model's page and not ranked. Prices are read every 3 days from 8 providers' public catalogs, against a typical $40 per 1M characters. Last refresh: 31 Oct 2026.